An Deep encoder and decoder convolutional neural network based retinal blood vessel segmentation
Abstract
Eye is an absolute sensory organ for vision in human. Vision in human is accomplished by blood vessels in retina and neurons in eye. Diseases such as diabetes retinopathy, hypertension and arteriosclerosis cause change in branching pattern and diameter of retinal blood vessels leading to blindness. These changes can be analyzed by segmenting retinal blood vessel. In this paper we propose two different architecture of neural network which is a deep learning based method to segment the retinal blood vessels. The feature map of fundus images are extracted by multiple hidden layers and are classified based on pixel classification. Loss function is performed to avoid losses due to imbalance between vessel and non vessel. The proposed work is evaluated for both architectures with DRIVE database.